A Digital Fountain Approach to Reliable Distribution of Bulk Data

A Digital Fountain Approach to Reliable Distribution of Bulk Data

© 1998 ACM 1-58113-003-1/98/0008 ... $15.00 | John W. Byers * Michael Luby† Michael Mitzenmacher‡ Ashutosh Rege §
The paper introduces a digital fountain approach to reliably distribute bulk data to a large number of autonomous clients. The digital fountain protocol allows any number of heterogeneous clients to acquire bulk data efficiently at their choosing, without the need for feedback channels to ensure reliable delivery, even in high-loss environments. The authors develop a protocol that closely approximates a digital fountain using a new class of erasure codes called Tornado codes, which are significantly faster than standard erasure codes for large block sizes. The paper provides performance measurements demonstrating the feasibility of the approach and discusses the design, implementation, and performance of an experimental system. The system is built on IP Multicast and uses layered multicast techniques to handle receiver heterogeneity and congestion control. The results show that Tornado codes outperform traditional Reed-Solomon codes in terms of encoding and decoding times, making them suitable for bulk data distribution over various network environments, including the Internet, satellite, and wireless networks.The paper introduces a digital fountain approach to reliably distribute bulk data to a large number of autonomous clients. The digital fountain protocol allows any number of heterogeneous clients to acquire bulk data efficiently at their choosing, without the need for feedback channels to ensure reliable delivery, even in high-loss environments. The authors develop a protocol that closely approximates a digital fountain using a new class of erasure codes called Tornado codes, which are significantly faster than standard erasure codes for large block sizes. The paper provides performance measurements demonstrating the feasibility of the approach and discusses the design, implementation, and performance of an experimental system. The system is built on IP Multicast and uses layered multicast techniques to handle receiver heterogeneity and congestion control. The results show that Tornado codes outperform traditional Reed-Solomon codes in terms of encoding and decoding times, making them suitable for bulk data distribution over various network environments, including the Internet, satellite, and wireless networks.
Reach us at info@study.space